Numerical modelling of ellipsoidal inclusions

Within the framework of numerical algorithms for the threedimensional random packing of granular materials this work presents an innovative formulation for polydispersed ellipsoidal particles, including an overlapping detection algorithm for an optimized simulation of the mesostructure of geomaterials, particularly concrete. Granular composite cement-based materials can be so reconstructed with adequate precision in terms of grain size distribution. Specifically, the algorithm performance towards the assumed inclusion shape (ellipsoidal or spheric) and degree of regularity (round or irregular) is here discussed. Examples on real grading curves prove that this approach is effective. The advantages of the proposed method for computational mechanics purposes are also disclosed when properly interfaced with visualization CAD (Computer Aided Design) tools.